Design of two-fuzzy neural-network controller for nonlinear systems

In this paper, we propose two-fuzzy neural-network controller (TFNNC) for nonlinear control systems. For the convenience of adaptive control, the structure of the two-fuzzy neural-network controller is divided into two parts. Each part is a fuzzy neural-network. Beginning, two-fuzzy neural-networks have the same structure. In the actual control process, at the same time, one fuzzy neural-network is used to be controller, its output as control input, and the other fuzzy neural-network is learning, the learning is the main on-line tracking learning. In fuzzy control rule table, we can get a switching line. By using the switching line, the actions of two fuzzy neural-networks can be exchanged. Different from other self-adaptive control laws, here, the process of self-adaptive and control is divided, i.e., the control is implemented by a fuzzy neural-network, and the self-adaptive is implemented by another fuzzy neural-network that is on-line tracking learning. This make the process of learning does not affect real-control. Simultaneously, the learning (self-adaptive) is on-line tracking. The stability analysis of this control law is given in the paper, and the conclusion shows that it is useful.